The study also had limited statistical power with a
finite number of observations in a limited number of
patients, thus fulfilling the criteria for a pilot investigation.
Pilot investigations are necessary to establish a prima
facie case for further extensive and expensive studies.
Multivariable logistic regression analysis was not
performed to enable us to confirm or refute the
hypothesis that chloride load is truly an independent
predictor of bicarbonate, base excess, and pH value. Such
analyses were precluded by two factors. There was
strong collinearity (variance inflation factor > 5) among
the biochemical variables making such multivariable
analysis statistically incorrect and there are known
physiological links among all the relevant biochemical
variables, such that a change in sodium or albumin will
have repercussion on bicarbonate, pH, base excess and
potassium. Thus, these biochemical variables are neither
statistically nor physiologically independent